With rapid development of Internet technologies, types of applications developed by developers become more, and applications used by users also become more. When a user uses an application, in order to quickly start up the application, a startup procedure of the application needs to be accelerated.
Currently, a startup accelerating procedure is specifically that: when a terminal monitors that the application is started up, a process in a startup procedure of the application is suspended. A disk file of the terminal is searched for corresponding pre-fetched information according to a name of the process and content data of the process, and the pre-fetched information is obtained by monitoring data read by the process from a magnetic disk within a preset period of time. If the pre-fetched information is found, the pre-fetched information is parsed, and a path, an offset value and a length value of data not loaded in a cache are obtained from the pre-fetched information. The corresponding data is pre-fetched according to the obtained path, offset value and length value, and the pre-fetched data is loaded into the cache. In this case, the process is restored, so that the process reads needed data from the cache, and if the cache does not have the needed data, the needed data may be read from the magnetic disk. A speed at which a process reads data from a cache is greater than a speed at which the data is read from a magnetic disk, and therefore startup accelerating is implemented.
Different process startup procedures consume different times, and therefore for different processes, when a terminal pre-fetches data according to pre-fetched information, only data in a preset period of time can be pre-fetched, and when the time consumed by the process startup procedure is less than the preset period of time, the terminal pre-fetches other data except needed data loaded by the process, or When the time consumed by the process startup procedure is greater than the preset period of time, the data pre-fetched by the terminal is less than the needed data loaded by the process. Moreover, when a process is started up, a terminal first suspends the process, and the process is restored after loading of pre-fetched data is completed, and if accuracy of the pre-fetched data is low, which results in a low hit rate of the pre-fetched data, a startup procedure of the process is prolonged.